Jakub Smékal comments on Sparse Autoencoders Work on Attention Layer Outputs

[Jakub Smékal]

Hey, great post! Are your code or autoencoder weights available somewhere?